Fix startup sequence to ensure backend readiness

- Update start.sh to wait for Unix socket creation before starting workers
- Remove retry logic from workers since startup script ensures backend is ready
- Workers now connect immediately knowing backend socket exists
- Prevents race conditions between backend startup and worker connections

start.sh

@@ -72,7 +72,27 @@ start_process() {
 # Start backend
 start_process "backend" "$PYTHON_CMD -m vidai.backend"
 
-$SLEEP_CMD 2
+# Wait for backend to be ready
+echo "Waiting for backend to be ready..."
+if [ $IS_WINDOWS -eq 1 ]; then
+    # On Windows, just sleep
+    $SLEEP_CMD 3
+else
+    # On Unix-like systems, check if socket exists
+    max_wait=10
+    for i in $(seq 1 $max_wait); do
+        if [ -S "/tmp/vidai_worker.sock" ]; then
+            echo "Backend is ready!"
+            break
+        fi
+        echo "Waiting for backend... ($i/$max_wait)"
+        $SLEEP_CMD 1
+        if [ $i -eq $max_wait ]; then
+            echo "Backend failed to start within $max_wait seconds"
+            exit 1
+        fi
+    done
+fi
 
 # Start workers with configured backends
 start_process "analysis_worker" "$PYTHON_CMD -c \"from vidai.worker_analysis import worker_process; worker_process('$ANALYSIS_BACKEND')\""